Accounting Discretion of Banks During a Financial Crisis

WP/09/207 Accountng Dscreton of Banks Durng a Fnancal Crss Harry Huznga and Luc Laeven

2009 Internatonal Monetary Fund WP/09/207 IMF Workng Paper Research Department Accountng dscreton of banks durng a fnancal crss Prepared by Harry Huznga and Luc Laeven 1 Authorzed for dstrbuton by Stjn Claessens September 2009 Abstract Ths Workng Paper should not be reported as representng the vews of the IMF. The vews expressed n ths Workng Paper are those of the author(s) and do not necessarly represent those of the IMF or IMF polcy. Workng Papers descrbe research n progress by the author(s) and are publshed to elct comments and to further debate. Ths paper shows that banks use accountng dscreton to overstate the value of dstressed assets. Banks balance sheets overvalue real estate-related assets compared to the market value of these assets, especally durng the U.S. mortgage crss. Share prces of banks wth large exposure to mortgage-backed securtes also react favorably to recent changes n accountng rules that relax far-value accountng, and these banks provson less for bad loans. Furthermore, dstressed banks use dscreton n the classfcaton of mortgage-backed securtes to nflate ther books. Our results ndcate that banks balance sheets offer a dstorted vew of the fnancal health of the banks. JEL Classfcaton Numbers: A10; A11 Keywords: Key bank regulaton, accountng standards, far value accountng, corporate dsclosure, fnancal crss Author s E-Mal Address: H.P.Huznga@uvt.nl; LLaeven@mf.org 1 Huznga s Professor of Economcs at Center, Tlburg Unversty, and Research Fellow at CEPR; Laeven s Senor Economst at the Internatonal Monetary Fund, and Research Fellow at CEPR. We would lke to thank Rocco Huang, Chrstan Leuz, Joe Mason, Lev Ratnovsk, and Wolf Wagner for comments or suggestons, and Matta Landon for excellent research assstance. The fndngs, nterpretatons, and conclusons expressed n ths paper are entrely those of the authors. They should not be attrbuted to the IMF. Contact nformaton: Harry Huznga: H.P.Huznga@uvt.nl; Luc Laeven: llaeven@mf.org.

2 Contents Page I. Introducton...3 II. Tobn s q Value and Market Dscounts...7 III. The Data...8 IV. Market Dscounts and Valuaton Effects of Real Estate Related Assets...12 A. Emprcal Evdence on Market Dscounts...12 B. Banks Stock Prce Reacton to Amendments of Far Value Accountng Rules...16 V. Accountng Dscreton on Impared Assets and Asset Classfcaton...18 A. Accountng Dscreton on Accountng for Bad Loans...18 B. Classfcaton of Mortgage-Backed Securtes...20 VI. Conclusons...21 References...24 Appendx Varable Defntons and Data Sources...29 Tables 1. Summary Statstcs for 2008, Quarterly Data... 30 2. Tobn s q and Real Estate Related Assets n 2008...31 3. Tobn s q and Real Estate Related Assets n 2001-2007...32 4. Tobn's q Real Estate Related Assets and Asset Sze...33 5. Tobn s q and Addtonal Balance Sheet and Off-balance Sheet Items...34 6. Event Study of New FASB Rules on Far Value Accountng for Illqud Assets (FAS 157), Announced on October 10, 2008...35 7. Event Study of FASB Amendments to Far Value Accountng of Hard-to-Value Assets, Announced on Aprl 9, 2009...36 8. Loan Loss Provsons and Net Loan Charge-offs n 2008...37 9. Share of Mortgage-backed Securtes that s Held-to-Maturty n 2008...38 10. Share of Non-Guaranteed Mortgage-backed Securtes that s Held-to-Maturty n 2001-2007...39 Fgures 1. Tobn s q and Share of Zombe Banks...40 2. Real Estate Loans and Mortgage-backed Securtes...40 3. Share of Mortgage-backed Securtes that s Held-to-Maturty...40 4. Far Value of Mortgage-backed Securtes Relatve to Amortzed Cost...41 5. Ter 1 Captal Rato and Share of Ter 1 Captal n Total Captal...41

3 I. INTRODUCTION The current fnancal crss has renvgorated a debate on the effectveness of the exstng accountng and regulatory frameworks for banks. Questons abound, rangng from adequate captalzaton levels of banks to the boundares of fnancal regulaton (see Fnancal Stablty Forum, 2008). Part of the debate on fnancal reform centers around requred nformaton on banks for effectve market dscplne and supervsory acton. Ths ncludes not only thnkng on the requred level of detal on dsclosure of bank assets and labltes but also on ther valuaton technques and the approprateness of current accountng rules more generally (see Laux and Leuz, 2009, for a survey). Part of ths debate centers around the pros and cons of far value accountng, where far value s meant to ndcate the prce at whch an asset could be bought or sold n a current transacton between wllng partes, other than n a lqudaton. Accountng standards stpulate that as a gudng prncple, the quoted market prce n an actve market should be used as the bass for the measurement of the far value of an asset. The problem s that such a prce s not always avalable, for example, n llqud markets. In such cases, far values need to be estmated based on avalable nformaton. A related concern s the potental procyclcal nature of far value accountng, whch could magnfy fluctuatons n bank lendng and economc actvty (see IMF, 2009, and Heaton et al., 2009). A broader concern s that the current mxed attrbute model of accountng, n whch some fnancal nstruments are measured based on hstorcal cost and some at far value, together wth dscreton over how fnancal nstruments are measured, gves rse to accountng arbtrage. 2 Despte dffcultes of determnng far values n llqud markets, advocates of far value accountng mantan that far value s the most relevant measure for fnancal nstruments. 3 They argue that fnancal assets, even complex nstruments, tend to trade contnuously n markets and t should therefore be possble to use nformaton embedded n market prces to compute far values of fnancal assets. Faced wth massve wrte-downs and expected losses, banks n contrast have used the momentum to lobby aganst the use of far value accountng. They clam that most of ther assets are currently not mpared, that they ntend to hold them to maturty anyway, and that market prces reflect dstressed sales nto an llqud market. Potental buyers of such assets, however, are unlkely to value them at orgnaton value but at prces well below book value. Banks may gnore such sgnals to avod recognzng a loss, clamng that unusual market condtons, not an actual declne n value, cause low market prcng. 2 As emphaszed by Jackson M. Day, Deputy Chef Accountant, U.S. Securtes and Exchange Commsson, n hs year 2000 remarks Far Value Accountng Let's Work Together and Get It Done! at the 28th Annual Natonal Conference on Current SEC Developments. 3 See, for example, Kaplan, Robert, Robert Merton and Scott Rchard, 2009, Dsclose the far value of complex securtes, Fnancal Tmes, August 17, 2009.

4 Accountng technques do not generally generate large dfferences between the book and market value of bank assets. At tmes of fnancal crss when asset markets become dstressed, however, large dfferences between book and market values of assets may arse, especally when assets are carred at values based on hstorcal cost. Such dfferences may gve rse to ncentves for banks to use accountng dscreton to preserve the book value of the bank, for example, by usng advantageous asset classfcatons or valuaton technques. As a consequence, dscreton n accountng rules enables banks to understate underlyng balance sheet stresses. Overstated book values of bank assets may further gve rse to undue regulatory forbearance. 4 Durng the ongong fnancal crss, large dfferences have ndeed arsen between market and book values of U.S. banks as ther market values have sharply eroded on the expectaton of major wrtedowns and losses on real estate related assets. By end-2008, more than 60% of U.S. bank holdng companes had a market-to-book value of assets of less than one, compared to only 8% of banks at end-2001. At the same tme, the average rato of Ter 1 captal to bank assets has stayed constant at about 11% throughout ths perod. The market value of bank equty thus has dropped precptously aganst a backdrop of vrtually constant book captal. Ths rases doubts about the relevance and relablty of banks accountng nformaton, the two man crtera on the bass of whch accountng systems are evaluated, at a tme of fnancal crss. Ths paper shows that banks use accountng dscreton to systematcally understate the mparment of ther real estate related assets, especally followng the onset of the current fnancal crss, n an effort to preserve book captal. We provde the frst evaluaton of such behavor and offer three peces of compellng evdence to support our thess that banks use accountng dscreton to overstate the book value of captal. Frst, we estmate large market dscounts on real estate related assets, ncludng mortgage loans and mortgage-backed securtes (MBS). To estmate mplct market dscounts on bank assets, we emprcally relate Tobn s q, computed as the market-to-book value of assets, to banks asset exposures usng quarterly accountng data on U.S. bank holdng companes for the perod 2001 to 2008. Our prmary focus s on real estate related assets, as these assets consttute a large fracton of the total assets of the average bank, and as recent declnes n U.S. real estate prces have rased doubts about the underlyng value of these assets. However, we also apply our methodology to other on- and off-balance sheet tems. We estmate sgnfcant dscounts on banks real estate loans (relatve to other loans) startng n 2005, averagng about 10% n 2008. As the average bank holdng company n 2008 holds about 54% of ts assets n the form of real estate loans, the mplct dscount n loan values goes a long way toward explanng the current depressed state of bank share prces. We further fnd that nvestors started dscountng banks holdngs of MBS n 2008. For that year, we fnd an average dscount on these assets of 24% (relatve to other securtes), whle the average MBS exposure amounted to 10% of assets. The market dscount on MBS that are avalable-for-sale (and carred at far value) s estmated to be 23%, aganst a dscount of 32% for MBS that are held-to-maturty (and carred at values based 4 For evdence of regulatory forbearance and the poltcal economy of bank nterventon, see Kane (1989), Kroszner and Strahan (1996), Barth et al. (2006), and Brown and Dnc (2005, 2009).

5 on hstorcal cost). Thus, even MBS that are carred at far value appear to be overvalued on the balance sheets of banks. Second, usng an event study methodology we fnd that banks wth large exposure to MBS experenced relatvely large excess returns when rules regardng far value accountng were relaxed. Pressures arose durng the summer of 2008 to provde banks wth more lenency to determne the far value of llqud assets such as thnly traded MBS to prevent these far values from reflectng fre sale prces. 5 Correspondngly, on October 10, 2008 the Fnancal Accountng Standards Board (FASB) clarfed the allowable use of non-market nformaton for determnng the far value of fnancal assets when the market for that asset s not actve. Subsequently, on Aprl 9, 2009, the FASB announced a related decson to provde banks greater dscreton n the use of non-market nformaton n determnng the far value of hard-to-value assets. As expected, the stock market on both occasons cheered the banks enhanced ablty to mantan accountng solvency n an envronment of low transacton prces for MBS. Usng an event study methodology, we fnd that banks wth large exposure to MBS experenced relatvely large excess returns around both announcement dates, ndcatng that these banks n partcular are expected to beneft from the expanded accountng dscreton. Thrd, we show that banks use accountng dscreton regardng the realzaton of loan losses and the classfcaton of assets to preserve book captal. Banks have consderable dscreton n the tmng of ther loan loss provsonng for bad loans and n the realzaton of loan losses n the form of charge-offs. Thus, banks wth large exposure to MBS and related losses can attempt to compensate by reducng the provsonng for bad debt n an effort to preserve book captal. We ndeed fnd that banks wth large portfolos of MBS report relatvely low rates of loan loss provsonng and loan charge-offs. We also examne banks choces regardng the classfcaton of MBS as ether held-tomaturty or avalable-for-sale. We consder ths categorzaton separately for MBS that are covered or ssued by a government agency. In 2008, the far value of especally non-guaranteed MBS tended to be less than ther amortzed cost. Ths mples that banks could augment the book value of assets by classfyng MBS as held-to-maturty. Indeed, we show that the share of nonguaranteed MBS that are held-to-maturty ncreased substantally n 2008. Classfcaton of ths knd s partcularly advantageous for banks whose share prce s depressed on account of large real estate related exposures. Consstent wth ths, we fnd that the share of MBS kept as held-tomaturty s sgnfcantly related to both real estate loan and MBS exposures. Moreover, these relatonshps are stronger for low-valuaton banks. Taken together, the evdence of ths paper shows that banks use consderable accountng dscreton regardng the categorzaton of assets, valuaton technques, and the treatment of loan losses. Accountng dscreton appears to be used to soften the mpact of the crss on the book 5 The prmary concern was one of mantanng solvency at affected banks. There was also a concern that losses nduced by fre sales could spread to other fnancal nsttutons. Allen and Carlett (2008) and Plantn et al. (2008) offer theoretcal models nvestgatng potental contagon effects among banks f far value accountng forces banks to value ther securtes accordng to observed fre sale prces.

6 valuaton of assets. Whle some accountng dscreton s unavodable as accountng systems n part are mechansms for frms to reveal asymmetrc nformaton to nvestors and other outsde partes 6, accountng dscreton entals the rsk of generatng hghly naccurate accountng nformaton at a tme of great turmol, such as the present fnancal crss. Inaccurate accountng nformaton n the case of banks can be especally harmful, as t may lead to regulatory forbearance wth concomtant rsks for tax payers. In the present crss, the fnancal statements of banks appear to overstate the book value of assets to the pont of becomng msleadng gudes to nvestors and regulators alke. 7 Thus, the present crss can be seen as a stress test of the accountng framework that reveals that book valuaton need not always reflect the best estmate of asset value, especally at a tme of sharp declnes n market values. Accountng reforms announced so far and dscussed n ths paper, however, seem to go n the drecton of ncreasng the gap between book and market values. Ths may be testmony that bank nterests wegh heavly n ths debate. Our paper relates to a large lterature n accountng and fnance on how accountng prncples and systems affect corporate behavor and that of banks n partcular (see, e.g., Collns et al., 1995, Shackelford et al., 2008, and Leuz and Wysock, 2008). Much of ths work analyzes the cost and benefts of earnngs management of frms (see, e.g., Leuz et al., 2003, and Hutton et al., 2008). There s also work on the costs and benefts of enhanced corporate dsclosure and accountng transparency (see Leuz and Wysock, 2008, for a revew). For example, Karpoff et al. (2008) usng frm-level nformaton on legal enforcement actons show that fnancal msrepresentaton has reputatonal consequences for frms and depresses frm valuaton. A related lterature revewed by Barth et al. (2001) and Holthausen and Watts (2001) asks whether accountng nformaton s value relevant n the sense that t conforms to the nformaton that bank shareholders use to prce bank shares. Barth et al. (1996) and Eccher et al. (1996) fnd that far value estmates of loan portfolos and securtes help to explan bank share prces beyond amortzed cost. There s also recent work on the market prcng of bank assets reported under dfferent far valuaton technques (e.g., Kolev, 2009, Goh et al., 2009, and Song et al., 2009). Bongn et al. (2002) further fnd that measures of bank fraglty based on market nformaton are a better predctor of bank falures than measures of bank fraglty based on accountng nformaton. Our paper s part of an emergng lterature on the causes and effects of the 2007 U.S. fnancal crss. Ths work shows that house prce apprecaton (e.g., Demyanyk and Van Hemert, 2008) and asset securtzaton (e.g., Keys et al., 2008; Man and Suf, 2008; Loutskna and Strahan, 2009), combned wth a more general deteroraton of lendng standards by banks (e.g., 6 A theoretcal lterature outlnes that managers of frms may have ncentves to smooth reported accountng ncomes ether to smooth ther own compensaton, to ncrease ther job securty, or to ncrease frm valuaton by nvestors (see, e.g., Trueman and Ttman, 1988, Fudenberg and Trole, 1995, and Sankar and Subramanyam, 2000). 7 The outcomes of stress tests of major U.S. banks conducted by the U.S. Treasury n 2009, whch calculated captal shortfalls at several major banks, are testmony to the fact that publcly avalable accountng nformaton at the tme provded an nadequate pcture of the health of the concerned banks.

7 Dell Arcca et al., 2008), helped fuel a crss n U.S. mortgage markets, wth bank captal beng eroded as the asset prce bubble n real estate markets burst startng n 2007. The paper contnues as follows. Secton II sets out the relatonshp between Tobn s q and market dscounts on bank assets. Secton 3 dscusses the data. Secton IV frst presents emprcal evdence on market dscounts of real estate related assets relatve to book values. Subsequently, t provdes evdence on the stock market response to the announcements of more lenent rules for accountng for llqud assets. Secton V examnes the use of bank dscreton regardng loan loss provsonng, loan charge-offs, and the classfcaton of MBS nto dfferent accountng categores. Secton VI concludes. II. TOBIN S Q VALUE AND MARKET DISCOUNTS In ths secton, we descrbe how observatons of Tobn s q can be used to nfer dscounts on bank assets mplct n the stock market. 8 Let MV be the market value of the bank. At the same tme, let A be the accountng value of asset and let L be the accountng value of lablty. Assumng there are operatng markets for a bank s assets and labltes, we can state a bank s market value as follows: MV v A v L (1) a l where a v s the market value of asset and l v s the market value of lablty. 9 We can defne q as the market value of the equty of the bank plus the book value of all labltes dvded by the book value of all assets as follows: q MV L A Substtutng for MV from (1) nto the expresson for q, we get: 8 In smlar fashon, Sachs and Huznga (1987) estmate dscounts on thrd world debt on the books of U.S. commercal banks at the tme of the nternatonal debt crss of the 1980s. A related lterature, startng wth Lang and Stulz (1994) and ncludng Laeven and Levne (2007), has studed dscounts n Tobn s q arsng from corporate dversfcaton. In that lterature, dscounts are computed for each busness unt of a conglomerate wth respect to the value of comparable stand-alone frms, whle here we compute dscounts for dfferent assets and labltes of the same bank. 9 In eq. (1), we gnore that market value may depend on the co-exstence of certan assets and labltes as dscussed n, for nstance, DeYoung and Yom (2008).

8 q 1 d a d l (2) a l where d a 1 v, d a l 1 v, a l A A and l L A. Note that a d and l d are the dscounts mplct n the bank s stock prce of a bank s assets and labltes relatve to book values. At the same tme, a and l are the accountng values of partcular assets and labltes relatve to the book value of all assets. From eq. (2), we see that f all assets and labltes of the bank are valued at market value n the bank s balance sheet, then q equals 1. Alternatvely, a devaton of q from 1 mples that the market valuaton of at least one balance sheet tems dffers from ts accountng value. 10 III. THE DATA In ths study, we consder U.S. bank holdng companes that are stock exchange lsted. These companes report a range of accountng data to the Federal Reserve System by way of the Report on condton and ncome (Call report). We are usng quarterly data from these Call reports from the fnal quarter of 2001 tll the end of 2008. Ths covers a full busness cycle as defned by the Natonal Bureau of Economc Research (NBER) from the prevous recesson whch ended n November 2001 untl the current ongong recesson whch started n December 2007. Our focus s on the year 2008, one year nto the recesson and what s generally consdered the start of the U.S. mortgage default crss (see for example Dell Arcca et al., 2008, and Man and Suf, 2008), when delnquences on mortgage loans ncreased sharply. Usng stock market data from Datastream, we use the market value of common equty plus the book value of preferred equty and labltes as a proxy for the market value of a bank s assets. Tobn s q s then constructed as the rato of ths proxy for the market value of bank assets and the book value of assets. Fgure 1 reports the average Tobn s q per quarter over our sample perod. The mean value of q has declned from 1.064 n the fnal quarter of 2001 to 0.998 n the fnal quarter of 2008. Ths suggests that over ths perod, the market value of bank assets has declned more than ts book value. We defne a zombe bank as a bank wth a q of less than one. 11 The declne of the average q has been accompaned by an ncrease of the share of banks that are zombe banks. As presented 10 Current book values of, say, real estate loans could already reflect some loan loss provsonng. Estmated dscounts on bank assets then reflect the dfference between market percepton of asset mparment and the recognton of ths mparment through reported loan loss provsonng (rather than the dfference between market value and orgnaton value). Put dfferently, the estmated dscount reflects the dfference between market percepton of any asset mparment and the accountng treatment of ths mparment. 11 The term zombe bank has frequently been used n the context of Japan durng the 1990 s bankng crss when Japanese banks contnued to lend to unproftable borrowers (e.g., Caballero et al., 2008).

9 n Fgure 1, the share of zombe banks has ncreased from 8.2% at the end of 2001 to 60.4% at the end of 2008. Durng ths perod, the share of zombe banks has tended to be smaller than n 2001 and 2008 reflectng an upswng of the busness cycle. In fact, the share of zombe banks reached a low of 0.3% durng the second quarter of 2004. U.S. banks are exposed to the real estate market n two mportant ways. Frst, they have sgnfcant portfolos of real estate loans. As an ndex of ths exposure, we construct the rato of real estate loans to overall assets. From 2001 to 2008 ths share of real estate loans has ncreased substantally from 45.2% to 53.3% for the average bank holdng company as reflected n Fgure 2. Thus, about half of the average bank s assets consst of real estate loans by 2008. In addton, banks are exposed to the real estate market through ther holdngs of MBS. Interestngly, the average rato of the book value of MBS to the book value of all assets has ncreased only slghtly from 10.0% n 2001 to 10.2% at the end of 2008. Whle there has been a move towards far value accountng of bank assets, most assets of the average bank, ncludng mortgage loans held for nvestment, are stll reported based on hstorcal cost. 12 The book value of MBS reflects dfferent accountng conventons dependng on whether these securtes are held-to-maturty or avalable-for-sale. MBS classfed as held-tomaturty are carred at amortzed cost. Ths amortzed cost may be adjusted perodcally for captalzed nterest and may also reflect prevous loan loss provsonng. However, these adjustments to amortzed cost are lkely to be relatvely small so that amortzed cost s relatvely close to orgnaton values. Alternatvely, MBS can be avalable-for-sale. In ths case, these securtes are to be carred at far value. Far value s meant to reflect observed market values (of ether the underlyng asset level 1 assets or a comparable asset level 2 assets) or otherwse reflect the outcome of a bank s own valuaton models (level 3 assets). 13 Agan, banks assessments of far value may dffer across bankng nsttutons as the determnaton of far value n practce leaves banks wth sgnfcant dscreton. 14 At any rate, at a tme of declnng asset values, one expects far values to be less than amortzed cost. 12 The majorty of (real estate) loans are carred at hstorcal cost, as loans held for sale, that are reported at the lower of hstorcal cost and far value, consttute only a small fracton of less than 1% of total assets for the average bank. 13 A breakdown of far value assets by valuaton technque (level 1 to 3) s n prncple avalable from Schedule HC- Q of the Call report. We do not use ths nformaton n our analyss, because, unlke securtes that are reported at both amortzed cost and far value, these assets are reported for only one of the three far valuaton technques, makng t dffcult to draw any nference based on a drect comparson of the amount of assets reported n each category. Furthermore, the level 1 to 3 assets are not broken down separately for real-estate related assets, whch are the prmary focus of our study, and are reported only for a subset of banks that have elected to report such assets under a far value opton. Moreover, the majorty of these assets are valued as level 2 assets (about 90 percent of far value assets n 2008), so there s not much varaton n far valuaton technque. 14 Indeed, work by Kolev (2009), Goh et al. (2009), and Song et al. (2009) shows that market dscounts dffer for level 1, level 2, and level 3 assets.

10 Interestngly, banks report n ther Call report flngs both the amortzed cost and far value of MBS regardless of whether these are held-to-maturty or avalable-for-sale. Thus, for MBS that are carred at amortzed cost we also know the assessed far value, whle for MBS carred at far value we also know the reported amortzed cost. Ths enables us to compute a bank s share of MBS that are held-to-maturty (rather than avalable-for-sale) on a sngle accountng bass. Specfcally, we can compute the share of MBS that s held-to-maturty usng amortzed costs for all MBS. The share of MBS that s held-to-maturty s computed separately for MBS that do and do not beneft from some explct or mplct offcal guarantee. Guaranteed MBS are those that are guaranteed or ssued by U.S. government agences such as the Federal Natonal Mortgage Assocaton (FNMA), the Federal Home Loan Mortgage Corporaton (FHLMC), and the Government Natonal Mortgage Assocaton (GNMA), more generally known as Fanne Mae, Fredde Mac, and Gnne Mae, respectvely. 15 Fgure 3 shows that for most of the sample perod the share of non-guaranteed MBS classfed as held-to-maturty exceeded the analogous share of guaranteed securtes. Moreover, durng 2008 the share of non-guaranteed MBS labeled held-tomaturty rose strongly from 8.3% to 11.7%. Durng that year, the share of guaranteed MBS that s held-to-maturty, nstead, fell from 6.5% to 6.0%. Classfcaton of MBS as held-to-maturty ncreases the book value of assets f far value s less than amortzed cost. Fgure 4 reports the mean rato of far value to amortzed cost as reported by dfferent banks over the sample perod separately for guaranteed and non-guaranteed MBS (regardless of whether these securtes are classfed as held-to-maturty or avalable-forsale). We see that ths rato s farly close to one for guaranteed MBS throughout the sample perod. For non-guaranteed MBS, however, far values relatve to amortzed cost declned from one n 2001 to 87.1% on average at end-2008. The ncreased classfcaton of non-guaranteed MBS as held-to-maturty durng 2008 (as seen n Fgure 3) has thus tended to boost the overall book value of banks MBS assets. Although the market value of most banks equty declned sharply n 2008, banks regulatory captal, as measured by the rato of Ter 1 captal to total rsk-weghted assets, has remaned farly stable throughout the sample perod. Fgure 5 shows the development of the Ter 1 captal rato and the share of Ter 1 captal n total bank captal. Whle leverage ncreased for some banks, consstent wth fndngs by Adran and Shn (2008), the average rato of Ter 1 captal to total assets decreased only modestly from 12.2% n 2001 to 11.1% n 2008. 16 The composton of captal also altered only modestly over the sample perod, wth the share of Ter 1 captal n total captal shrnkng from 88.2% n 2001 to 86.3% n 2008. Ths suggests that, although some banks have looked for less tradtonal, non-core sources of captal, such as 15 Note that these guarantees tend to cover underlyng repayment of nterest and prncple, but not valuaton rsk stemmng from nterest rate changes or mortgage prepayment. 16 Ter 1 captal represents the core component of captal for banks and s regarded as the key measure of a bank s fnancal strength from a regulator s pont of vew. Ter 1 captal conssts prmarly of common stock, retaned earnngs, and dsclosed reserves.

11 subordnated debt or perpetual stock, to boost captal and ncrease assets, most banks contnued to do so whle ncreasng Ter 1 captal and mantanng excess regulatory captal. A bank s q should be close to one n a world where all bank assets and labltes are readly tradable and marked to market. At the same tme, devatons of q from one can be explaned by dscrepances between market values and book values of any bank balance sheet tems. Below, we relate a bank s q to a range of bank balance sheet tems to explan bank-level varaton n q. Varable market values of bank balance sheet tems n an envronment of slowly adjustng book values suggest that the dependence of q on bank balance sheet tems vares over tme. It s especally nterestng to assess whether the valuaton of bank balance sheet tems mplct n bank stock prces dffers from book values at a tme of fnancal crss. Therefore, the emphass of the emprcal work wll be on the year 2008, the year followng the onset of the U.S. mortgage default crss. Summary statstcs for the man varables n 2008 are provded n Table 1. We exclude banks wth Tobn s q exceedng ts 99 th percentle (amountng to a Tobn s q greater than 1.5) as these are not ordnary banks that carry prmarly fnancal assets. The mean rato of loans to assets s 71.4%, whle the mean rato of real estate loans to assets s 53.6%. The rato of securtes to assets (usng amortzed cost to value held-to-maturty securtes and far values for securtes avalable-for-sale) s 16.9%. As a subcategory, the average rato of MBS to assets s 9.6%. Ths can be splt nto MBS held-to-maturty at 0.8% of assets, and MBS avalable-for-sale at 8.8% of assets. MBS that are held-to-maturty can agan be splt nto guaranteed and nonguaranteed securtes equvalent to 0.7% and 0.1% of assets, respectvely. Guaranteed and nonguaranteed MBS that are avalable-for-sale n turn amount to 8.0% and 0.8% of assets. Next, Large bank s a dummy varable that equals one f a bank s total assets exceed the sample average total assets n a gven quarter, and zero otherwse. HPI s a state-level house prce ndex from the U.S. Offce of Federal Housng Enterprse Oversght (OFHEO). Low valuaton s a dummy varable that equals one n a gven quarter f a bank s q s less than one, and zero otherwse. By the end of 2008, 60% of U.S. banks had a value of q of less than one. Several addtonal asset categores are consdered as well. Tradng s defned as tradng assets relatve to total assets (obtaned from Schedule HC-B of the Call report). Tradng assets, whch nclude some MBS, are carred at far value and held n the bank s tradng book. 17 A detaled splt-up of tradng assets s only avalable for the domestc offces of bank holdng companes and s not reported. On average, tradng assets only amount to a share of 0.5% of assets, because only large banks tend to have such assets. Among bank lablty varables, Deposts s defned as total deposts dvded by total assets, and t amounts to 72.0% of assets on average. These deposts nclude relatvely stable retal deposts and more unstable wholesale deposts. Data on deposts are obtaned from Schedule HC-E of the Call report fles. As an ndex of unstable wholesale deposts, we construct 17 Tradng assets are to be reported only by bank holdng companes wth average tradng assets of $2 mllon or more n any of the four precedng quarters.

12 the rato of deposts that exceed $ 100,000 and have a remanng maturty of less than one year to total assets. These large and short-term deposts on average are 2.5% of assets. Banks are further seen to ssue relatvely lttle commercal paper, wth commercal paper amountng to only 0.l% of assets on average. Bank captal, beng the sum of Ter 1 and Ter 2 captal, s composed mostly of Ter 1 captal, amountng to 86.3% of captal on average. Off-balance sheet tems can equally matter for bank valuaton. However, we fnd that they tend to consttute a small fracton of total assets for the average bank, n part because only large banks tend to have sgnfcant off-balance exposure. 18 Data on off-balance sheet tems are obtaned from Schedule HC-L of the Call report fles. Credt dervatves postve and Credt dervatves negatve are the mean ratos of credt protecton purchased and credt protecton sold to total assets, respectvely. These ratos are equvalent to 1.5% and 1.4% of assets. We also obtan nformaton on banks securtzaton and asset sale actvtes from Schedule HC-S of the Call report fles. The varable Securtzed s the rato of assets sold and securtzed wth servcng retaned by the bank, or wth recourse or other seller provded credt enhancements, to total assets. Securtzed takes on a value of 1.5% of assets on average. Asset sales stands for the rato of assets sold wth recourse or other seller-provded credt enhancements but not securtzed to total assets, and takes on a mean value of 3.2% of assets. The mean values of these off-balance sheet tems are seen to be small on average and they are expected to affect bank valuaton correspondngly lttle. Next, loan loss provsonng s calculated as loan loss provsons dvded by the book value of all loans. The mean loan loss provsonng rate s 0.8%. Net charge-offs, n turn, s the rato of the dfference between loan charge-offs and loan recoveres to the book value of loans. The mean net loan charge-off rate s 0.5%. Thus, loan loss provsonng exceeded net loan charge-offs n 2008, as expectatons of addtonal loan losses surpassed actual loan wrte-offs. Fnally, the share of real estate loans s the rato of real estate loans to total loans wth a mean value of 74.3%. IV. MARKET DISCOUNTS AND VALUATION EFFECTS OF REAL ESTATE RELATED ASSETS Ths secton frst provdes emprcal estmates of market dscounts of real estate related assets relatve to book values. Subsequently, t examnes bank stock prce reactons to amendments of far value accountng rules. Fnally, t nvestgates the use of banks dscreton regardng the accountng for bad loans n the form of loan loss provsonng and loan charge-offs. A. Emprcal Evdence on Market Dscounts Ths subsecton reports the results of regressons of q to reveal mplct stock market valuatons of key balance sheet and off-balance sheet tems. All regressons nclude U.S. state 18 It should also be noted that only banks wth off-balance sheet exposures n excess of certan mnmum values are requred to report these exposures.

13 fxed effects and quarterly perod fxed effects to control for systematc dfferences across U.S. states and tme perods, such as housng and labor market condtons, or the monetary polcy stance. To start, Table 2 reports regressons of q that nclude the overall loans and overall securtes varables wth data for 2008. 19 The Securtes varable enters wth a postve coeffcent of 0.096, whch suggests that overall securtes are valued more hghly mplct n bank share prces than on banks books, though the effects s not statstcally sgnfcant. The Loans varable also does not enter sgnfcantly. Next, regresson 2 n addton ncludes the real estate loans and MBS varables. Note that the ncluson of both the Real estate loans varable and the Loans varable, whch ncludes real estate loans, mples that the effect of real estate loans s measured relatve to that of other loans. Smlarly, for MBS, the effect s computed relatve to the overall effect for Securtes, snce MBS are a part of total securtes. The real estate loans varable enters wth a coeffcent of -0.107 that s sgnfcant at the 1% level mplyng that the dscount of real estate loans (relatve to other loans) s 10.7%. The drect effect of real estate loans on Tobn s q, computed by addng the coeffcents of the Loans and Real estate loans varables, s close to zero, ndcatng that non-real estate loans carry a negatve dscount. The MBS varable smlarly enters wth a coeffcent of - 0.244 that s sgnfcant at the 1% level so that MBS appear to be dscounted 24.4% relatve to other securtes. 20 In regresson 3, we replace the MBS varable wth two separate varables, MBS, held and MBS, for sale that represent the parts of MBS that are held-to-maturty (and carred at amortzed cost) and avalable-for-sale (and carred at far value). The MBS, held varable obtans a coeffcent of -0.321 that s sgnfcant at 1%, whle the MBS, for sale varable enters wth a coeffcent of -0.227 that s sgnfcant at 5%. Thus, MBS classfed as held-to-maturty appear to be dscounted sgnfcantly at 32.1%, whle the MBS avalable-for-sale tend to have a smaller dscount of 22.7% on average relatve to other securtes. Thus, the gap between mplct market prces and accountng values appears to be largest for MBS classfed as held-to-maturty. Fnally, n regresson 4 we splt the MBS, held and MBS, for sale varables nto ther guaranteed and non-guaranteed parts. Now we see that the guaranteed and non-guaranteed parts of the MBS, held varable are estmated wth coeffcents of -0.293 and -0.472 that are both sgnfcant at the 1% level, whle the two MBS, for sale varables obtan negatve coeffcents of -0.220 and -0.324 that are smaller n absolute value. Thus, especally the non-guaranteed MBS 19 The estmaton model mplctly sets the dscount on excluded asset categores to zero. Asset categores excluded from the regresson are cash-lke assets, ncludng cash and federal funds sold and amountng to 9% of total assets, and non-cash lke assets, ncludng tradng assets and fxed assets and amountng to the remander of 2% of total assets. Thus, wth cash-lke assets carryng a dscount of close to zero and consttutng the majorty of excluded assets, the mplct assumpton of a dscount of zero on excluded asset categores appears to be reasonable. 20 We only consder the market valuaton of MBS as mplct n share prces. Emprcal models of the drect prcng of MBS are offered by Dunn and Sngleton (1983), Boudoukh et al. (1997), and Schwartz and Torous (1989).

14 classfed as held-to-maturty are dscounted (relatve to securtes other than MBS). The mpled dscount of 47.2% for these non-guaranteed MBS s szeable. The evdence thus ponts at szeable market dscounts on real-estate related assets relatve to book values for U.S. bank holdng companes n 2008. As we have data from 2001 onward, t s nterestng to see whether such dscounts exsted before 2008. For ths purpose, we re-estmate regresson 3 of Table 2 wth data for each of the years n the perod 2001-2007. The results are reported n Table 3. Throughout the perod 2001 to 2004, none of the real estate asset categores s estmated wth a sgnfcant dscount. From 2005, the real estate loan varable obtans ncreasngly negatve coeffcents of -0.074, -0.081 and -0.101 that are sgnfcant at the 1% level to ndcate a gradual deteroraton of the mplct market value of real estate loans relatve to book value. The MBS varables, however, are not estmated wth sgnfcant dscounts throughout the 2001-2007 perod. The deteroraton of real estate loans thus appears to have preceded the deteroraton of MBS by several years, untl n 2008 both asset categores are estmated wth sgnfcant dscounts. We want to make sure that our results are not entrely drven by the sze of the bank. To ths end, we re-estmate regresson 3 of Table 2 separately for small and large banks by splttng the sample based on the Large bank varable. The results are reported n regressons 1 and 2 of Table 4. Except for the nfluence of Loans and Real estate loans, we fnd lttle dfference n the estmated coeffcents of the real-estate related varables for small and large banks. The dscount on real estate loans s estmated to be 15.1% and sgnfcant at the 1% level for large banks, whle the dscount s estmated to be nsgnfcant for small banks. At the same tme, non-real estate loans are estmated to carry a premum for small banks. The estmated coeffcent on the MBS, held varable s somewhat more negatve for small banks, although ths varable enters wth statstcally sgnfcant coeffcents for both small and large banks. The estmated coeffcent on the MBS, for sale varable s more negatve for large banks than for small banks. In regresson 3 we agan consder the full sample of banks but nclude nteracton terms of the real estate loans and MBS varables wth the Large bank dummy varable that denotes whether the bank s large or small. The regresson confrms that the nfluence of real estate loans s statstcally dfferent for large banks compared to small banks. Specfcally, the dscount on real estate assets s estmated to be 12.7% larger for large banks. The estmated dscounts for the MBS varables, on the other hand, turn out not to be statstcally sgnfcantly dfferent between small and large banks. So far, we have focused on loans and securtes and ther real estate components. Ths emphass s justfed by the fact that loans and securtes together comprse on average 88.3% of bank assets n 2008, and by the fact that real estate assets have suffered from house prce declnes durng the recent fnancal crss. Nevertheless, t s nterestng to nclude other on- and off-balance sheet tems n the analyss as well.

15 To start, regresson 1 of Table 5 ncludes several addtonal asset categores n regresson 4 of Table 2. We splt the MBS varables nto ther guaranteed and non-guaranteed parts but ths does not affect the results. The regressons results ndcate that non-guaranteed and held-tomaturty MBS are dscounted the most. Tradng, denotng the rato of tradng assets to total assets, enters the two regressons wth negatve but nsgnfcant coeffcents. The mprecse estmaton of the coeffcent on the tradng varable could reflect that tradng assets, n fact, nclude many dverse assets and on average comprse only 0.5% of total assets n 2008. Next, regresson 2 of Table 5 ncludes several lablty varables. Frst, Deposts stands for the rato of total deposts to total assets. We expect ths varable to carry a postve coeffcent because banks extract value from the government guarantee on deposts n the presence of depost nsurance that s ncreasng n the amount of deposts. Indeed, we fnd that ths varable obtans a postve though nsgnfcant coeffcent. Second, Deposts, large, short-term stands for the rato of deposts n excess of $100,000 and wth a remanng maturty of one year or less to total assets. These large and short-term deposts can be consdered part of the wholesale fundng of a bank. The supply of ths type of bank fundng may be unstable, not least because deposts n excess of $100,000 are tradtonally not covered by depost nsurance. Ths varable enters wth a coeffcent of -0.157 that s sgnfcant at the 5% level. Ths suggests that 1 dollar of these wholesale deposts reduces bank value by about 0.16 dollars (more than other deposts). Ths, of course, does not mean that the market value of these deposts s substantally dfferent from unty. Rather, a bank that heavly reles on wholesale fundng s exposed to consderable fundng rsk as potentally reflected n bank share prces. Thrd, the commercal paper varable stands for the rato of ssued commercal paper to total assets. Ths varable enters wth a postve but nsgnfcant coeffcent. Regresson 3 ncludes a varable that captures the composton of equty captal. Specfcally, we nclude the share of Ter 1 captal n total captal, denoted by the Ter 1 varable. We expect that ths varable enters wth a postve coeffcent, especally for the year 2008, as markets have reassessed the superor value of Ter 1 captal to Ter 2 captal, partly n response to strcter captal requrements proposed by regulators. We ndeed fnd that the Ter 1 captal varable enter wth postve coeffcents of 0.105 that s sgnfcant at the 1% level. Ths suggests that a one standard devaton ncrease of 10% n the share of Ter 1 captal n total captal ncreases bank value by 1%, whch s not rrelevant gven a standard devaton of q of 5%. Interestngly, n unreported regressons we fnd that pror to 2008 the effect of the share of Ter 1 captal on q s not statstcally sgnfcant, ndcatng that Ter 1 or core captal became a hghly valued component of bank captal only startng n 2008. We next nclude several off-balance sheet tems n regresson 4, ncludng nformaton on credt protecton purchased or sold, asset securtzaton, and asset sales. 21 None of these offbalance sheet varables enter sgnfcantly, possbly reflectng the fact that they consttute only a small fracton of bank s assets. 21 The varables n the expresson for q reman defned as shares of the value of on-balance sheet assets.

16 Comparng the results of regresson 4 n Table 2 wth those of regresson 4 n Table 5, we see that the ncluson of addtonal balance sheet varables reduces estmated coeffcents for the loans and securtes varables and renders these varables nsgnfcant. Thus, the mplct stock market valuaton of non-real estate loans and securtes does not dffer sgnfcantly from book valuaton n Table 5. Real estate related varables, however, contnue to obtan negatve and sgnfcant coeffcents. The negatve and sgnfcant coeffcent on the real estate loan varable mples that real estate loans are dscounted relatve to non-real estate loans as well as relatve to book values. Smlarly, MBS that are held-to-maturty and avalable-for-sale are dscounted relatve to non-mbs securtes and relatve to book values. One concern s that our results are drven by an overshootng n asset prces, meanng a temporary devaton n value from fundamental value. However, our measure of frm value s based on equty prces, whch reflect the consensus vew of many fnancal market partcpants. Whle fre sales and llqudty may have led to overshootng n some asset markets, notably the market for dervatves on mortgage-backed securtes, stock markets contnued to be lqud throughout 2008. We therefore mantan that stock market prces offer the best avalable nformaton on the value of banks, and conclude that the accountng values of real estate related assets on the books of banks were nflated n 2008. B. Banks Stock Prce Reacton to Amendments of Far Value Accountng Rules Thus far, we have studed the mpact of banks asset composton on the valuaton of banks to gauge the market dscounts mplct n dfferent assets. Dfferences n such market dscounts partly reflect dfferences n accountng treatment. In ths secton, we assess how recent changes to accountng rules have affected the valuaton of banks by studyng the mmedate stock prce reacton to the announcements of these rule changes. On October 10, 2008, the FASB clarfed rules for determnng the far value of a fnancal nstrument applyng Fnancal Accountng Standard (FAS) 157 when the market for that fnancal asset s not actve. 22 The clarfcaton made explct that the use of a bank s own assumptons about future cash flows and approprately rsk-adjusted dscount rates s acceptable when relevant observable nputs nto value calculaton are not avalable. Also, t was made clear that broker (or prcng servce) quotes may be approprate nput when measurng far value. 23 These announced nterpretatons of FAS 157 were seen to provde banks wth more dscreton n determnng the far value of securtes and to enable them to lmt markdowns n the face of llqud securtes markets durng the U.S. mortgage default crss. 22 These rules, ssued under Fnal Staff Poston on FAS 157-3, were effectve upon ssuance, ncludng pror perods for whch fnancal statements have not been ssued. 23 The Offce of the Chef Accountant of the U.S. Securtes and Exchange Commsson (SEC) and the FASB staff had already jontly ssued a press release on September 30, 2008, that addresses smlar applcaton ssues of FAS 157. See http://www.fasb.org/news/2008-farvalue.pdf for further detals.

17 Subsequently, on Aprl 9, 2009 the FASB approved amendments to FAS 157 that gve banks more dscreton n usng non-market nformaton to determne far values of securtes. 24 In practce, frms wll be allowed to re-classfy level 2 assets, whch were prevously valued usng proxy reference market prces, to level 3 assets, whose valuaton s model-based. 25 By provdng greater flexblty n excludng llqud transactons from level 2 far value determnaton, the new rules effectvely expand the scope for frms to prevent sgnfcant mark-downs n llqud markets subject to great prce declnes, and possbly to mark-up assets that had been aggressvely wrtten down prevously. 26 Both the October 2008 and Aprl 2009 announcements of the FASB were seen by market commentators as efforts to artfcally prop up the accountng value of banks. Whle resultng n a decrease n transparency and nformaton dsclosure, these changes are expected to be cheered by shareholders of dstressed banks, because the reducton n wrtedowns allows such banks to mantan regulatory captal requrements. We use a standard event study methodology to compute the average prce effect on bank shares of these announcements of changes n accountng rules. Also, we assess whether the share prces of dfferent types of banks reacted dfferently to these announcements. In partcular, we examne whether abnormal returns vary by bank sze and the degree to whch banks hold MBS. We use a standard market model to estmate abnormal returns. Table 6 reports the event study results for the October 10, 2008 announcement. Cumulatve abnormal returns are based on a market model wth estmaton wndow of [t-250, t- 30], where t denotes October 10, 2008, and tme s counted n tradng days. We use the total return on the S&P 500 as proxy for the daly market return. We report results for two dfferent event wndows. Panel A reports results usng an event wndow of (t-3, t+2], where t denotes October 10, 2008, and tme s counted n tradng days, whle Panel B reports results usng an event wndow of (t-1, t]. Usng such a short event wndow of a sngle day s acceptable gven the hgh stock market volatlty around the tme of ths event, culmnatng n a stock market crash. To mtgate concerns that returns from llqud frms are drvng the result, we exclude from the sample observatons from frms wth more than 100 zero returns over the estmaton wndow or a zero return on the event date. Average cumulatve abnormal returns are reported both for the full sample of banks and for dfferent subsamples of banks, wth sample splts based on a host of bank characterstcs, specfcally bank sze and the degree to whch banks hold MBS. The sample splts are as follows: 24 The changes became effectve for fnancal statements endng June 2009, wth early adopton permtted for frstquarter 2009 results. 25 See http://www.fasb.org/news/nr040909.shtml for further detals. 26 On the same day, new accountng rules were announced that wll reduce the level of losses to be dsclosed n frms ncome statements for avalable-for-sale and held-to-maturty debt securtes. Under the old rules, provded the frm had the ntent and ablty to hold the securty untl recovery, other-than-temporary mparment would need to be recognzed n the ncome statement. Under the new rules, provded the frm does not have the ntent to sell the securty, t only needs to recognze the credt component of the other-than-temporary mparment n ncome, whle recordng the remanng porton n a specal category of equty ( other comprehensve ncome ). The change from ntent and ablty to hold to no ntenton to sell may provde suffcent flexblty to sgnfcantly reduce the level of total mparment, of whch only the credt component s deducted from ncome.

18 Large (small) denotes frms wth total assets above (below) the quarterly sample medan; and Hgh (Low) share of MBS denotes frms wth MBS as a fracton of total assets above (below) the quarterly sample medan. We use thrd quarter 2008 Call report data to construct these bankspecfc varables, whle daly total return data on equtes are obtaned from Datastream. Table 7 reports event study results for the second event on Aprl 9, 2009. Agan, we report results separately for two dfferent event wndows n Panels A and B. To avod valuaton effects arsng from events that occurred durng the perod followng the announcement of the frst event, ncludng the frst event tself, from basng the market model nduced estmates of normal returns, we apply the same estmaton wndow as used n the frst event study to estmate normal returns. The cumulatve abnormal returns (CAR) are large on the event day tself for both events but n the case of the frst event, the average CAR across all banks s much lower and barely sgnfcant f we extend the event wndow. The reason s that October 10, 2008 was the only day that week durng whch the stock market experenced postve returns n what otherwse was a rapdly fallng market, n whch the prces of bank stocks were fallng more sharply than those of non-bank stocks. The sample splts reveal a number of nterestng dfferences n the valuaton effect across dfferent types of banks. The CAR of large banks s consstently hgher and economcally large. One explanaton for ths result s that larger U.S. banks tend to have a larger fracton of hard-tovalue assets, ncludng off-balance sheet, and thus tend to beneft most from the changes n accountng rules. The share prce of banks wth a large fracton of MBS also reacts favorably to the relaxaton of far value accountng, at least for the October 10, 2008 event, as expected. Overall, we fnd that the valuaton of large banks and banks wth a large fracton of MBS gans relatvely much on account of both announcements. Ths can be explaned by the fact that these banks have relatvely many assets such as MBS that are affected by more lenent rules regardng the calculaton of ther far value. V. ACCOUNTING DISCRETION ON IMPAIRED ASSETS AND ASSET CLASSIFICATION In ths secton, we assess the relevance of banks dscreton n accountng for bad loans and n classfyng MBS nto categores that render more favorable accountng values. Together wth the valuaton results presented n secton 4, these results shed lght on the relablty of banks fnancal statements, and n partcular on the extent to whch book values of banks assets accurately account for future asset mparment. A. Accountng dscreton on accountng for bad loans The relatve mportance of real estate assets n the average bank s portfolo renders bank captal very senstve to the performance of real estate loans. In case of expected future loan losses, a bank needs to provson for these losses. Provsonng for loan losses, however, reduces

19 ncome and regulatory captal. Thus, dstressed banks may be tempted to provson relatvely less for real estate loans or any other loans n an attempt to overstate captal. 27 In ths subsecton, we report regressons that test whether dstressed banks report relatvely low loan loss provsons. To capture loan loss provsonng, we construct the rato of loan loss provsons to total loans. 28 We obtan data on loan charge-offs and provsons from Schedule HI-B of the Call report fles. In regresson 1 of Table 8, the loan loss provsonng varable s frst related to the share of real estate loans n total loans. We expect loan loss provsonng to be postvely related to the share of real estate loans, as these loans have been partcularly affected by recent house prce declnes. The share of real estate loans ndeed enters the regresson wth a postve coeffcent, but t s statstcally nsgnfcant. Banks that need to absorb large losses arsng from exposure to MBS may lower ther provsonng standards n an effort to preserve captal. As a proxy for potental losses arsng from exposure to MBS, we use the rato of MBS to assets denoted MBS. Ths exposure varable obtans a negatve coeffcent of -0.015 that s statstcally sgnfcant at the 5% level, suggestng that banks wth large MBS exposure tend to attenuate reported loan loss provsons. We expect that the ncentve to hold back on loan loss provsonng s partcularly pronounced for dstressed banks. Regressons 2 and 3 therefore re-estmate regresson 1 for subsamples of banks wth below-medan and above-medan q, respectvely. Regresson 2 confrms a negatve and statstcally sgnfcant coeffcent for the MBS varable for lowvaluaton banks, whle the coeffcent for the MBS varable s negatve but nsgnfcant n regresson 3. Thus, low-valuaton banks appear to be the ones that compensate for ther MBS exposure by scalng back ther loan loss provsonng. Dstressed banks also may be slow n recognzng losses on ther real estate loan portfolo n the form of wrte-downs 29 or charge-offs. 30 To analyze ths, regressons 4 to 6 take as dependent varable the rato of net charge-offs to loans (where net charge-offs are the dfference between charge-offs and recoveres). Otherwse, these regressons are smlar to regressons 1 to 3. Consstent wth the earler results, we now fnd that the rato of net charge-offs to loans s negatvely related to the MBS varable, though the effect s not statstcally sgnfcant. 27 Prevously, Moyer (1990) and Ahmed et al. (1999) have found that banks use ther dscreton regardng loan loss provsonng to manage ther captal. Dockng et al. (1997) consder the nformaton and contagon effects of bank loan loss reserve announcements. 28 No breakdown of loan loss provsonng for real estate loans and other loan categores s avalable from banks Call reports. 29 Loan wrtedowns nclude wrtedowns arsng from transfers of loans to a held-for-sale account. 30 Loan charge-offs reduce allowances for loan losses rather than bank captal f prevous loan loss provsons were made. In any case, charge-offs may trgger further loan loss provsonng whch reduces regulatory captal.

20 In sum, we fnd evdence that low-valuaton banks wth large MBS exposures hold back on ther loan loss provsonng. B. Classfcaton of Mortgage-Backed Securtes Accordng to FAS 159, banks have the opton to classfy securtes as held-to-maturty or avalable-for-sale. Securtes are to be classfed as held-to-maturty and carred at amortzed cost, f management has the ntenton to hold them untl maturty. Otherwse, securtes are avalable-for-sale and carred at far value. Ths classfcaton s to be made on the date of purchase of the securty and t s n prncple rreversble. However, banks can acheve some reclassfcaton of prevously acqured securtes n complance wth FAS 159 by sellng and buyng equvalent securtes that are categorzed dfferently wthn the same reportng perod. On the purchase date, amortzed cost and far value should be essentally the same and hence no valuaton advantage can be obtaned by classfyng securtes ether way. 31 Reclassfcaton of prevously acqured securtes potentally does affect the overall book value of securtes. Specfcally, overall book value rses f avalable-for-sale securtes are reclassfed as held-to-maturty at a tme when amortzed cost exceeds far value. In 2008, the mean rato of far value to amortzed cost for non-guaranteed MBS was 0.927, aganst a mean rato of far value to amortzed cost for guaranteed MBS of 1.005 (see Fgure 4). These accountng valuatons gave banks an ncentve to classfy non-guaranteed MBS as held-to-maturty to the extent possble. We now examne whether banks, and especally dstressed banks, responded to ths ncentve by classfyng a larger fracton of ther MBS as held-to-maturty. Table 9 reports regressons of the shares of MBS that are held-to-maturty for guaranteed as well as non-guaranteed securtes. In the calculaton of these shares, the MBS that are actually avalable-for-sale are also valued at amortzed cost. The number of observatons dffers dependng on whether the dependent varable s computed for guaranteed or non-guaranteed securtes because a sgnfcant fracton of banks reports not to have any non-guaranteed MBS. The fracton of MBS that s held-to-maturty ncreased from 7.5% at end-2007 to 11.7% at end- 2008, consstent wth the noton that banks had ncentves durng the year 2008 to classfy a larger fracton of ther MBS as held-to-maturty (see Fgure 3). The regresson 1 results ndcate that the share of guaranteed MBS classfed as held-tomaturty s postvely but nsgnfcantly related to both the real estate loans and the overall MBS (valued at amortzed cost) to assets varables. In regresson 2, we see that the non-guaranteed share of MBS that s held-to-maturty s postvely and nsgnfcantly related to the MBS, amortzed cost varable but postvely and sgnfcantly to the real estate loans varable wth a coeffcent of 0.731. 31 A consderaton gudng ths classfcaton at the tme of securtes acquston can be to obtan an approprate mx of assets and labltes that are carred at far value.

21 Thus, we fnd evdence that banks pressured by real estate exposure tend to report a relatvely large share of non-guaranteed MBS as held-to-maturty, and that ths effect operated chefly through exposure to real estate loans rather than MBS. Regressons 3 and 4 dffer from regressons 1 and 2 n that we nclude the Low valuaton varable as an addtonal varable to assess dfferences n the classfcaton of MBS between banks wth hgh or low q. The Low valuaton varables enters both regressons wth a postve but nsgnfcant coeffcent, ndcatng that there s no sgnfcant dfference between hgh and low valuaton banks n the fracton of non-guaranteed MBS that they report as held-to-maturty. Fnally, regressons 5 and 6 dffer from regressons 3 and 4 n that we nclude nteracton terms of the real estate exposure varables and the Low valuaton varable. Postve estmated coeffcents mply that especally banks wth below-average q report a larger share of ther MBS as held-to-maturty n response to large real estate exposures. Indeed, the nteracton terms n regressons 5 and 6 all enter wth postve estmated coeffcents, although the coeffcents are statstcally sgnfcant only for the nteracton wth the MBS varable n regresson 6. Ths suggests that banks wth below-average q ncrease the share of non-guaranteed MBS that s heldto-maturty to a relatvely large extent n response to real estate exposures. Ths s to be expected as the gans n terms of the book value of assets are relatvely large n the case of non-guaranteed MBS, as for these securtes the rato of far value to amortzed cost was relatvely low n 2008. Next, we examne whether banks have also exploted dscreton n the classfcaton of ther MBS wth a vew to boost the accountng value of ther assets pror to 2008. To do ths, we re-estmate regresson 4 of Table 9 wth data for the perod 2001-2007. A focus on nonguaranteed MBS s justfed, as the rato of far value to amortzed cost of these MBS devates relatvely frequently from unty as seen n Fgure 4. In 2001, for nstance, far values of nonguaranteed MBS tended to exceed amortzed cost. The results are presented n Table 10. The MBS, amortzed varable enters the regressons n Table 11 wth ether negatve or postve coeffcent, dependng on the year, although none of these estmated coeffcents s statstcally sgnfcantly dfferent from zero. The real estate loan varable enters wth postve coeffcents that are sgnfcant at the 5% level from the year 2005 onwards, suggestng that banks wth large real estate exposure classfed a larger fracton of ther non-guaranteed MBS as held-to-maturty. Over the 2002-2008 perod, the real estate loans varable ncreases n a nonmonotonc way from 0.137 to 0.457. Turnng to the Low valuaton varable, we fnd that ths varable enters wth postve but nsgnfcant coeffcents for the years 2001 through 2004 that turn negatve from the year 2005 onwards. Overall, these results confrm that already pror to 2008 banks classfed ther nonguaranteed MBS wth a vew to boastng the book value of these assets. VI. CONCLUSIONS In 2008, the majorty of U.S. banks were zombe banks as evdenced by market values of bank assets beng lower than ther book values. Ths s prma face evdence that the book value of banks balance sheets s nflated. We fnd that the stock market attaches less value to real

22 estate loans and MBS than ther accountng values. Ths dscrepancy between the accountng and market value of bank assets suggests that banks have been slow to adjust the book value of ther assets to conform to market expectatons about future declnes n asset performance. We further fnd a larger dscount for held-to-matury MBS (that are carred at amortzed cost) than for avalable-for-sale MBS (that are carred at far value), suggestng that far values recognze the mparment of MBS to a greater extent than amortzed costs do. We estmate valuatons mplct n bank share prces for a range of bank labltes and off-balance sheet tems as well. Bank share prces are found to negatvely reflect bank fundng n the form of large and short-term deposts. Ths may reflect that wholesale fundng of ths type exposes the bank to consderable fundng rsks. Bank share prces are further found to be affected by off-balance sheet tems such as credt nsurance bought and sold, as well as credt commtments to own and other fnancal nsttutons securtzaton structures. Whle we do not drectly address the ssue of procyclcalty of far value accountng, we fnd that at a tme of depressed asset prces such as n 2008, one year nto the U.S. mortgage crss, the stock market apples dscounts to banks that are larger than those mplct n the far values of MBS. Ths suggests that far value accountng, as currently mplemented, s stll less procyclcal than any accountng based exclusvely on stock market valuatons would be. 32 In October 2008 and Aprl 2009, the FASB announced sets of accountng rule amendments provdng banks wth addtonal dscreton n the determnaton of far value of securtes n case markets are llqud and transacton prces may result from fre sales. On both occasons, banks wth large exposures to MBS are found to have experenced relatvely large excess returns. Addtonal dscreton n the determnaton of far values n an envronment of depressed asset prces makes t easer for banks wth large affected exposures to mantan accountng solvency, whch s apparently cheered by bank equty nvestors. Ths paper further demonstrates that banks wth large exposures to MBS systematcally use ther accountng dscreton so as to nflate asset values and book captal. Specfcally, banks wth large exposure to MBS are found to report relatvely low loan loss provsonng rates and loan charge-off rates, and at the same tme they tend to classfy a relatvely large share of ther MBS as held-to-maturty, to be able to carry these assets at amortzed costs. Our fndng that dstressed banks tend to explot ther dscreton n loan loss provsonng, loan charge-offs, and classfcaton of MBS to boost ther accountng value should be reason for concern, as t mples that the dscreton mplct n current accountng rules leads to systematc bases n valuatons on bank balance sheets. Accountng dscreton enables banks wth mpared 32 At any rate, n our vew the man task of accountng systems s to provde relable nformaton, and ths goal should not be compromsed by concerns about any procyclcalty of credt supply. A common vew s that bank regulaton should target any undesrable credt procyclcalty drectly, for nstance by prescrbng cyclcal captal requrements (for a more detaled dscusson of ths debate, see Laeven and Majnon, 2003, Kashyap and Sten, 2004, and Repullo and Suarez, 2008).

23 asset portfolos to satsfy captal adequacy requrements, but t makes t dffcult to assess the true health of the affected banks. Replacng the mxed attrbute model of accountng wth a model based entrely on far value accountng wll mtgate ncentves for accountng arbtrage and could serve to mprove the nformaton value of publc accounts, even f far value calculatons themselves are also subject to dscreton by banks. Smlarly, a more forward-lookng approach to provsonng for bad loans on an expected loss bass could mprove the nformaton content of bank accountng, although ncentves for banks to use dscreton on loan loss provsonng rates to nflate the book value of assets durng economc downturns would reman. No accountng system of dsclosng the far value of fnancal assets wll be perfect. Models can be msused or msnterpreted. But reasonable and audtable methods exst today to ncorporate nformaton embedded n market prces. More relable publc accounts are benefcal to regulatory and market dscplne and could potentally have helped to avod some of the losses that banks currently face.

24 REFERENCES Adran, Tobas, and Hyun S. Shn, 2008, Lqudty and Leverage, Journal of Fnancal Intermedaton, forthcomng. Ahmed, Anwer S., Carolyn Takeda, and Shawn Thomas, 1999, Bank Loan Loss Provsons: A Reexamnaton of Captal Management, Earnngs Management and Sgnalng Effect, Journal of Accountng and Economcs Vol. 28, pp. 1-25. Allen, Frankln, and Elena Carlett, 2008, Mark-to-Market Accountng and Lqudty Prcng, Journal of Accountng and Economcs Vol. 45, pp. 358-78. Barth, James R., Gerard Capro, and Ross Levne, 2006, Rethnkng Bank Regulaton, Cambrdge: Cambrdge Unversty Press. Barth, Mary E., Wllam H. Beaver, and Wayne R. Landsman, 1996, Value-relevance of Banks Far Value Dsclosures Under SFAS, No. 107, Accountng Revew, pp. 513-37., 2001, The Relevance of the Value Relevance Lterature for Fnancal Accountng Standard Settng: Another Vew, Journal of Accountng and Economcs Vol. 31, pp. 77-104 Berger, Allen, Sally M. Daves, and Mark J. Flannery, 2000, Comparng Market and Regulatory Assessments of Bank Performance: Who Knows What When? Journal of Money, Credt and Bankng, Vol. 32 No. 3, pp. 641-67. Bongn, P., L. Laeven, and G. Majnon, 2002, How Good s the Market at Aassessng Bank Fraglty? A Horse race between dfferent ndcators, Journal of Bankng and Fnance, Vol. 26, pp. 1011 28. Boudoukh, Jacob, Matthew Rchardson, Rchard Stanton, and Robert F. Whtelaw, 1997, Prcng Mortgage-backed Securtes n a Multfactor Interest Rate Envronment: A Multvarate Densty Estmaton Aapproach, Revew of Fnancal Studes Vol. 10, pp. 405-46. Brown, Crag O., and I. Serdar Dnc, 2005, The Poltcs of Bank Falures: Evdence From Emergng Markets, Quarterly Journal of Economcs, Vol. 120 No. 4, pp. 1413-444., 2009, Too Many to Fal? Evdence of Regulatory Forbearance When the Bankng Sector s Weak, Revew of Fnancal Studes, forthcomng. Caballero, Rcardo, Takeo Hosh, and Anl K. Kashyap, 2008, Zombe Lendng and Depressed Restructurng n Japan, Amercan Economc Revew Vol. 98, pp. 1943-977.

25 Collns, Jule H., Douglas A. Shackelford, and James M. Wahlen, 1995, Bank Dfferences n the Coordnaton of Regulatory Captal, Earnngs, and Taxes, Journal of Accountng Research, Vol. 33 No. 2, pp. 263-91. Dell Arcca, Govann, Denz Igan, and Luc Laeven, 2008, Credt Booms and Lendng Standards: Evdence From the Subprme Mortgage Market, mmeo (Washngton: Internatonal Monetary Fund). Demyanyk, Yulya, and Otto Van Hemert, 2008, Understandng the Subprme Mortgage Crss, Revew of Fnancal Studes, forthcomng. DeYoung, Robert and Chwon Yom, 2008, On the Independence of Assets and Labltes: Evdence From U.S. Commercal Banks, 1990-2005, Journal of Fnancal Stablty Vol. 4, pp. 275-303. Dockng, Dane Scott, Mark Hrschey, and Elane Jones, 1997, Informaton and Contagon Effects of Bank Loan-loss Reserve Announcements, Journal of Fnancal Economcs Vol. 43, pp. 219-39. Dunn, Kenneth B., and Kenneth J. Sngleton, 1983, An Emprcal Analyss of the Prcng of Mortgage-backed Securtes, Journal of Fnance Vol. 38, pp. 613-23. Eccher, Elzabeth A., K. Ramesh, and S. Ramu Thagarajan, 1996, Far Value Dsclosures by Bank Holdng Companes, Journal of Accountng and Economcs Vol. 22, pp. 79-117. Fnancal Stablty Forum, 2008, Report on Enhancng Market and Insttutonal Reslence, Aprl 2008. Avalable at: http://www.fnancalstabltyboard.org/ Frexas, Xaver, and Dmtros Tsomocos, 2004, Book vs. Far Value Accountng n Bankng, and Intertemporal Smoothng, Oxford Fnancal Research Centre workng paper. Fudenberg, Drew, and Jean Trole, 1995, A Theory of Income and Dvdend Smoothng Based on Incumbency Rents, Journal of Poltcal Economy, Vol. 103, pp. 75-93. Goh, Beng Wee, Jeffrey Ng, and Kevn Ow Yong, 2009, Market Prcng of Banks Far Value Assets Reported Under SFAS 157 Durng the 2008 Economc Crss, Mmeo, MIT. Heaton, Johan, Deborah Lucas, and Robert McDonald, 2009, Is Mark-to-Markng Accountng Destablzng? An Analyss and Implcatons for Polcy, mmeo, Unversty of Chcago. Holthausen, Robert W., and Ross L. Watts, 2001, The Relevance of the Value-relevance Lterature for Fnancal accountng Standard Settng, Journal of Accountng and Economcs 2001, pp. 3-75. Hutton, Amy P., Alan J. Marcus, and Hassan Tehranan, 2008, Opaque Fnancal Reports, R2, and the Dstrbuton of Crash Rsks, Journal of Fnancal Economcs, forthcomng.

26 Internatonal Monetary Fund, 2008, Global Fnancal Stablty Report. Chapter 3: Far Value Accountng and Procyclcalty, October 2008. Kane, Edward J., 1989, The S&L Insurance Mess: How Dd It Happen? Washngton, D.C.: Urban Insttute., 2003, Contnung Dangers of Dsnformaton n Corporate Accountng Reports, Revew of Fnancal Economcs, Vol. 13, pp. 149-64. Karpoff, Jonathan M., D. Scott Lee, and Gerald S. Martn, 2008, The Cost to Frms of Cookng the Books, Journal of Fnancal and Quanttatve Analyss Vol. 43, pp. 581 612. Kashyap, Anl, and Jeremy Sten, 2004, Cyclcal Implcatons of the Basel II Captal Standards, Federal Reserve Bank of Chcago Economcs Perspectvesm 1 st quarter, pp. 18-31. Keys, Benjamn, Tanmoy Mukherjee, Amt Seru, and Vkrant Vg, 2008, Securtzaton and Screenng: Evdence From Subprme Mortgage Backed Securtes, Quarterly Journal of Economcs, forthcomng. Knaup, Martn, and Wolf Wagner, 2009, A Market-based Measure of Credt Qualty and Banks Performance Durng the Subprme Crss, mmeo, Tlburg Unversty. Kolev, Kaln, 2009, Do Investors Perceve Markng-to-Model as Markng-to-Myth? Early Evdence from FAS 157 Dsclosure, mmeo, NYU Stern School of Busness. Kroszner, Randall S., and Phlp E. Strahan, 1996, Regulatory Incentves and the Thrft Crss: Dvdends, Mutual-to-Stock Conversons, and Fnancal Dstress, Journal of Fnance, Vol. 51, pp. 1285 1319. Laeven, Luc, and Govann Majnon, 2003, Loan Loss Provsonng and Economc Slowdowns: too Much, too Late, Journal of Fnancal Intermedaton Vol. 12, pp. 178-97. Laeven, Luc, and Ross Levne, 2007, Is There a Dversfcaton Dscount n Fnancal Conglomerates? Journal of Fnancal Economcs Vol. 85, pp. 331-67. Landsman, Wayne, 2006, Far Value Accountng for Fnancal Instruments: Some Implcatons for Bank Regulaton, BIS Workng Papers No 2009. Lang, Larry, and Rene M. Stulz, 1994, Tobn s q, Corporate Dversfcaton, and Frm Performance, Journal of Poltcal Economy Vol. 102, pp. 1248-280. Laux, Chrstan, and Chrstan Leuz, 2009, The Crss of Far Value Accountng: Makng Sense of the Recent Debate, mmeo, Goethe Unversty Frankfurt.

27 Leuz, Chrstan, D.J. Nanda, and Peter Wysock, 2003, Investor Protecton and Earnngs Management: An Internatonal Comparson, Journal of Fnancal Economcs, Vol. 69, pp. 505-27. Leuz, Chrstan, and Peter Wysock, 2008, Economc Consequences of Fnancal Reportng and Dsclosure Regulaton: A Revew and Suggestons for Future Research, mmeo, Chcago Unversty and MIT. Loutskna, Elena, and Phlp Strahan, 2009, Securtzaton and the Declnng Impact of Bank Fnance on Loan Supply: Evdence From Mortgage Orgnatons, Journal of Fnance, Vol. 64, pp. 861-89. Man, Atf, and Amr Suf, 2008, The Consequences of Mortgage Credt Expanson: Evdence from the 2007 Mortgage Default Crss, Quarterly Journal of Economcs, forthcomng. Morgan, Donald P., 2002, Ratng Banks: Rsk and Uncertanty n an Opaque Industry, Amercan Economc Revew Vol. 92, pp. 874-88. Moyer, Susan E., 1990, Captal Adequacy and Rato Regulatons and Accountng Choces n Commercal Banks, Journal of Accountng and Economcs Vol. 13, pp. 123-54. Plantn, Gullaume, Haresh Sapra, and Hyun S. Shn, 2008, Markng-to-Market: Panacea or Pandora s box? Journal of Accountng Research Vol. 46, pp. 435-60. Repullo, Rafael, and Javer Suarez, 2008, The Procyclcal Effects of Basel II, CEPR Dscusson paper No. 6862, CEPR, London. Sachs, Jeffrey, and Harry Huznga, 1987, U.S. Commercal Banks and the Developng Country Debt Crss, Brookngs Papers on Economc Actvty Vol. 2, pp. 555-606. Sankar, Mandra R., and K. R. Subramanyam, 2000, Reportng Dscreton and Prvate Informaton Communcaton Through Earnngs, Journal of Accountng Research Vol. 39, pp. 365-86. Schwartz, Eduardo S., and Walter N. Torous, 1989, Prepayment and the Valuaton of Mortgage-backed Securtes, Journal of Fnance Vol. 44, pp. 375-92. Shackelford, Douglas A., Joel B. Slemrod, and James M. Sallee, 2008, A Unfyng Model of How the Tax System and Generally Accepted Accountng Prncples Affect Corporate Behavor, mmeo, Unversty of North Carolna at Chapel Hll. Song, Chang Joon, Wayne Thomas, and Han Y, 2009, Value Relevance of FAS 157 Far Value Herarchy Informaton and the Impact of Corporate Governance Mechansms, mmeo, Unversty of Oklahoma.

Trueman, Brett, and Sherdan D. Ttman, 1988, An Explanaton for Accountng Income Smoothng, Journal of Accountng Research, Vol. 26, pp. 127-39. 28

29 Appendx. Varable Defntons and Data Sources Varable Defnton Source Tobn s q Rato of market value of common equty plus book value of preferred equty and labltes to book value of assets Call report and Datastream Share of MBS held-tomaturty, Share of guaranteed mortgage-backed securtes (MBS) that s held-to-maturty Call report guaranteed Share of MBS held-tomaturty, Share of non guaranteed MBS that s held-to-maturty Call report not guaranteed Loans Rato of loans to assets Call report Real estate loans Rato of real estate loans to assets Call report Securtes Rato of securtes to assets. Securtes held-to-maturty are at amortzed cost and Call report securtes avalable-for-sale are at far value Securtes, amortzed cost Rato of securtes to assets. Securtes are at amortzed cost f they are both heldto-maturty Call report and avalable-for-sale MBS Rato of MBS to assets. Held-to-maturty securtes are at amortzed cost and Call report avalable-for-sale securtes are at far value MBS, amortzed Rato of MBS to assets. Both held-to-maturty and avalable-for-sale MBS are at Call report amortzed cost MBS, held Rato of MBS that are held-to-maturty to assets Call report MBS, for sale Rato of MBS that are avalable-for-sale to assets Call report MBS, held, guaranteed Rato of MBS that are held-to-maturty and ssued or guaranteed by FNMA, Call report FHLMC, and GNMA to assets MBS, held, not Rato of non-guaranteed MBS that are held-to-maturty to assets Call report guaranteed MBS, for sale, guaranteed Rato of MBS that are avalable-for-sale and ssued or guaranteed by FNMA, Call report FHLMC, and GNMA to assets MBS, for sale, not Rato of non-guaranteed MBS that are avalable-for-sale to assets Call report guaranteed Large bank Dummy varable that s one f assets are above mean of assets n the data set and Call report zero otherwse HPI State-level housng prce ndex, rescaled to ndex value of 1 OFHEO Low valuaton Dummy varable that equals 1 f Tobn s q s less than 1, and 0 otherwse Call report Tradng Rato of assets n tradng account to total assets Call report Deposts Rato of deposts to assets Call report Deposts, large, shortterm Rato of tme deposts of $100,000 or more wth a remanng maturty of one year Call report or less to assets Commercal paper Rato of commercal paper to assets Call report Ter 1 Rato of ter 1 captal n total captal Call report Credt dervatves, Rato of notonal amount of credt dervatves for whch the bank s the Call report postve benefcary (credt protecton purchased) to assets Credt dervatves, Rato of notonal amount of credt dervatves for whch the bank s the guarantor Call report negatve (credt protecton extended) to assets Securtzed Rato of outstandng prncpal balance of assets sold and securtzed wth servcng Call report retaned or wth recourse or other seller-provded credt enhancements to assets Asset sales Rato of assets sold wth recourse or other seller-provded credt enhancements Call report and not securtzed to assets Loan loss provsonng Rato of loan loss provsonng to loans Call report Net charge-offs Rato of loan charge-offs mnus recoveres to loans Call report Share of real estate loans Share of real estate loans n total loans Call report

30 Table 1. Summary Statstcs for 2008, Quarterly Data Mean St. dev. Mnmum Maxmum Number Tobn s q 1.0133 0.0550 0.8976 1.3280 1152 Share of MBS held-tomaturty, guaranteed 0.0628 0.1844 0 1 1118 Share of MBS held-tomaturty, not guaranteed 0.0957 0.2501 0 1 598 Loans 0.7142 0.1170 0.0517 0.9593 1152 Real estate loans 0.5360 0.1453 0 0.8880 1152 Securtes 0.1686 0.0987 0 0.7702 1152 MBS 0.0959 0.0777 0 0.5758 1152 MBS, held 0.0083 0.0319 0 0.3594 1152 MBS, for sale 0.0876 0.0699 0 0.4456 1152 MBS, held, guaranteed 0.0071 0.0289 0 0.3577 1152 MBS, held, not guaranteed 0.0012 0.0113 0 0.2006 1152 MBS, for sale, guaranteed 0.0798 0.0642 0 0.4009 1152 MBS, for sale, not guaranteed 0.0078 0.0174 0 0.1592 1152 Large bank 0.4983 0.5002 0 1 1152 HPI 3.9383 1.2065 2.0982 6.9824 1132 Low valuaton 0.4991 0.5002 0 1 1152 Tradng 0.0053 0.0253 0 0.2996 1152 Deposts 0.7194 0.1088 0.1227 0.9028 1152 Deposts, large, short-term 0.0250 0.0317 0 0.3580 1152 Commercal paper 0.0010 0.0052 0 0.0628 1152 Ter 1 0.8634 0.0816 0.5000 1 1152 Credt dervatves, postve 0.0147 0.1571 0 2.9420 1152 Credt dervatves, negatve 0.0142 0.1515 0 2.8203 1152 Securtzed 0.0145 0.0756 0 0.7292 1152 Asset sales 0.0032 0.0130 0 0.1440 1152 Loan loss provsonng 0.0077 0.0107-0.0004 0.1435 1152 Net charge-offs 0.0050 0.0077-0.0009 0.0906 1152 Share of real estate loans 0.7429 0.1480 0 1 1152 Note: See the appendx for varable defntons and data sources.

Table 2. Tobn s q and Real Estate Related Assets n 2008 (1) (2) (3) (4) Loans -0.038 0.101* 0.097* 0.090* (0.052) (0.053) (0.054) (0.054) Real estate loans -0.107*** -0.103*** -0.101*** (0.031) (0.031) (0.031) Securtes 0.096 0.277*** 0.278*** 0.274*** (0.060) (0.084) (0.085) (0.085) MBS -0.244*** (0.088) MBS, held -0.321*** (0.086) MBS, for sale -0.227** (0.095) MBS, held, guaranteed -0.293*** (0.089) MBS, held, not guaranteed -0.472*** (0.105) MBS, for sale, guaranteed -0.220** (0.096) MBS, for sale, not guaranteed -0.324* (0.195) Large bank 0.007 0.008 0.008 0.008 (0.006) (0.006) (0.006) (0.006) HPI 0.000 0.001 0.001 0.000 (0.007) (0.007) (0.006) (0.007) Constant 1.025*** 0.954*** 0.956*** 0.963*** (0.044) (0.041) (0.041) (0.042) 31 N 1132 1132 1132 1132 R 2 0.317 0.373 0.375 0.377 The dependent varable s Tobn s q. See the appendx for varable defntons and data sources. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons durng the year 2008. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely.

Table 3. Tobn s q and Real Estate Related Assets n 2001-2007 2001 2002 2003 2004 2005 2006 2007 (1) (2) (3) (4) (5) (6) (7) Loans 0.197 0.127*** 0.063 0.114** 0.166*** 0.124* 0.085 (0.136) (0.047) (0.053) (0.050) (0.052) (0.065) (0.064) Real estate loans -0.070-0.005 0.016-0.041-0.074** -0.081** -0.101*** (0.048) (0.035) (0.034) (0.034) (0.030) (0.033) (0.033) Securtes 0.131 0.076 0.063 0.069 0.092* 0.077 0.077 (0.121) (0.054) (0.055) (0.051) (0.051) (0.065) (0.068) MBS, held -0.041 0.066 0.017-0.034-0.069-0.049-0.086 (0.087) (0.080) (0.085) (0.063) (0.058) (0.069) (0.080) MBS, for sale -0.088-0.032-0.041 0.001-0.016-0.037-0.089 (0.072) (0.057) (0.048) (0.049) (0.050) (0.057) (0.068) Large bank 0.063*** 0.057*** 0.034*** 0.034*** 0.020*** 0.019*** 0.009* (0.009) (0.007) (0.006) (0.006) (0.005) (0.006) (0.005) HPI 0.022-0.024 0.008 0.013* 0.006 0.005 0.035*** (0.014) (0.023) (0.014) (0.007) (0.008) (0.017) (0.010) Constant 0.832*** 0.972*** 0.990*** 0.978*** 0.962*** 1.001*** 0.947*** (0.116) (0.057) (0.052) (0.039) (0.040) (0.067) (0.054) 32 N 286 1186 1250 1274 1297 1169 1172 R 2 0.371 0.337 0.257 0.311 0.327 0.316 0.354 The dependent varable s Tobn s q. See the Appendx for varable defntons and data sources. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons over the perod 2001-2007. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely.

Table 4. Tobn s q, Real Estate Related Assets, and Asset Sze Small banks Large banks Interactons wth Large bank (1) (2) (3) Loans 0.233*** 0.117 0.146** (0.078) (0.071) (0.058) Real estate loans -0.053-0.151*** -0.035 (0.037) (0.045) (0.036) Real estate loans * Large bank -0.127** (0.053) Securtes 0.375*** 0.394*** 0.346*** (0.077) (0.133) (0.091) MBS, held -0.409*** -0.274* -0.433*** (0.107) (0.151) (0.104) MBS, held * Large bank 0.210 (0.135) MBS, for sale -0.222** -0.270* -0.280*** (0.095) (0.143) (0.100) MBS, for sale * Large bank 0.070 (0.096) Large bank 0.071* (0.037) HPI -0.001 0.006 0.003 (0.010) (0.008) (0.006) Constant 0.832*** 0.972*** 0.874*** (0.060) (0.051) (0.052) 33 N 578 554 1132 R 2 0.408 0.495 0.403 The dependent varable s Tobn s q. See the Appendx for varable defntons and data sources. Subsample n Column (1) conssts of banks wth below-medan total assets n a gven quarter. Subsample n Column (2) conssts of banks wth above-medan total assets n a gven quarter. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons durng the year 2008. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely.

Table 5. Tobn s q and Addtonal Balance Sheet and Off-Balance Sheet Items (1) (2) (3) (4) Loans 0.035 0.018 0.019 0.018 (0.080) (0.084) (0.079) (0.078) Real estate Loans -0.102*** -0.092*** -0.100*** -0.101*** (0.031) (0.031) (0.031) (0.031) Securtes 0.206* 0.191* 0.141 0.135 (0.114) (0.115) (0.117) (0.116) MBS, held, guaranteed -0.286*** -0.270*** -0.260*** -0.257*** (0.089) (0.090) (0.089) (0.088) MBS, held, not guaranteed -0.448*** -0.420*** -0.389*** -0.387*** (0.109) (0.107) (0.106) (0.110) MBS, for sale, guaranteed -0.207** -0.194** -0.174* -0.171* (0.098) (0.096) (0.095) (0.096) MBS, for sale, not guaranteed -0.329* -0.286-0.187-0.182 (0.194) (0.207) (0.211) (0.213) Tradng -0.190-0.168-0.142-0.039 (0.135) (0.166) (0.157) (0.181) Deposts 0.044 0.032 0.029 (0.052) (0.051) (0.053) Deposts, large, short-term -0.157** -0.161** -0.167** (0.079) (0.078) (0.079) Commercal paper 0.506 0.604* 0.675** (0.324) (0.317) (0.337) Ter 1 0.104*** 0.101*** (0.033) (0.034) Credt dervatves, postve -0.082 (0.258) Credt dervatves, negatve 0.064 (0.260) Securtzed -0.019 (0.034) Asset sales 0.003 (0.007) Large bank 0.008 0.009 0.012* 0.906*** (0.006) (0.007) (0.006) (0.078) HPI 0.001 0.003 0.003 0.018 (0.007) (0.007) (0.007) (0.078) Constant 1.008*** 0.976*** 0.898*** -0.101*** (0.064) (0.084) (0.077) (0.031) 34 N 1132 1132 1132 1132 R 2 0.380 0.389 0.401 0.403 The dependent varable s Tobn s q. See the Appendx for varable defntons and data sources. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons durng the year 2008. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely.

Table 6. Event Study of New FASB Rules on Far Value Accountng of Illqud Assets (FAS 157), Announced on October 10, 2008 Panel A: Event wndow s October 8, 2008 untl October 12, 2008 (1) (2) (3) (4) (5) All frms Large Small Low share of MBS Hgh share of MBS CAR 0.0128* 0.0260*** -0.0005 0.0116 0.0140 (0.0070) (0.0092) (0.0105) (0.0111) (0.0087) Observatons 270 136 134 134 136 Panel B: Event wndow s October 10, 2008 (1) (2) (3) (4) (5) All frms Large Small Low share of MBS Hgh share of MBS CAR 0.0761*** 0.1290*** 0.0225** 0.0616*** 0.0903*** (0.0074) (0.0079) (0.0107) (0.0113) (0.0094) Observatons 270 136 134 134 136 35 Ths table reports average cumulatve abnormal returns for dfferent subsamples of frms. Cumulatve abnormal returns are based on a market model wth estmaton wndow of [t-250, t-30], where t denotes October 10, 2008, and tme s counted n tradng days. Panel A reports results usng an event wndow of (t-3, t+2], where t denotes October 10, 2008, and tme s counted n tradng days, whle Panel B reports results usng an event wndow of (t-1, t]. Observatons from frms wth more than 100 zero returns over the estmaton wndow or a zero return on the event date are excluded from the sample. Large (small) denotes frms wth total assets above (below) the quarterly sample medan. Hgh (Low) share of MBS denotes frms wth mortgage-backed securtes as a fracton of total assets above (below) the quarterly sample medan. Standard errors of the average cumulatve abnormal returns are reported n parentheses. ***,**, and * denote sgnfcance at the 1%, 5%, and 10% level, respectvely.

Table 7. Event Study of FASB Amendments to Far Value Accountng of Hard-to-Value Assets, Announced on Aprl 9, 2009. Panel A: Event wndow s Aprl 7, 2009 untl Aprl 11, 2009 (1) (2) (3) (4) (5) All frms Large Small Low share of MBS Hgh share of MBS CAR 0.0643*** 0.0899*** 0.0390*** 0.0632*** 0.0654*** (0.0066) (0.0092) (0.0090) (0.0094) (0.0093) Observatons 255 127 128 127 128 Panel B: Event wndow s Aprl 9, 2009 (1) (2) (3) (4) (5) All frms Large Small Low share of MBS Hgh share of MBS CAR 0.0499*** 0.0662*** 0.0337*** 0.0497*** 0.0501*** (0.0043) (0.0054) (0.0063) (0.0067) (0.0053) Observatons 255 127 128 127 128 Ths table reports average cumulatve abnormal returns for dfferent subsamples of frms. Cumulatve abnormal returns are based on a market model wth estmaton wndow of [t-250, t-30], where t denotes October 10, 2008, and tme s counted n tradng days. Panel A reports results usng an event wndow of (t-3, t+2], where t denotes Aprl 9, 2009, and tme s counted n tradng days, whle Panel B reports results usng an event wndow of (t-1, t]. Observatons from frms wth more than 100 zero returns over the estmaton wndow or a zero return on the event date are excluded from the sample. Large (small) denotes frms wth total assets above (below) the quarterly sample medan. Hgh (Low) share of MBS denotes frms wth mortgage-backed securtes as a fracton of total assets above (below) the quarterly sample medan. Standard errors of the average cumulatve abnormal returns are reported n parentheses. ***,**, and * denote sgnfcance at the 1%, 5%, and 10% level, respectvely. 36

Table 8. Loan Loss Provsons and Net Loan Charge-offs n 2008 Loan loss provsonng Net loan charge-offs All banks Low valuaton Hgh valuaton All banks Low valuaton Hgh valuaton (1) (2) (3) (4) (5) (6) Share of real estate loans 0.004 0.000 0.000 0.003-0.001 0.000 (0.005) (0.008) (0.002) (0.003) (0.005) (0.002) MBS -0.015** -0.024* -0.002-0.008-0.010-0.000 (0.007) (0.013) (0.005) (0.005) (0.009) (0.004) Large bank 0.004*** 0.008*** 0.001* 0.003*** 0.005*** 0.001 (0.001) (0.002) (0.001) (0.001) (0.001) (0.000) HPI -0.005** -0.001-0.009*** -0.005** -0.003-0.006** (0.002) (0.003) (0.003) (0.002) (0.003) (0.003) Constant 0.013* 0.000 0.030*** 0.012* 0.008 0.019*** (0.008) (0.012) (0.007) (0.006) (0.010) (0.007) N 1132 562 570 1132 562 570 R 2 0.344 0.440 0.413 0.310 0.421 0.305 The dependent varable s the rato of loan loss provsonng to loans n Columns (1) to (3) and the rato of loan charge-offs mnus recoveres to loans n Columns (4) to (6). See the Appendx for varable defntons and data sources. Subsamples n Columns (2) and (5) consst of banks wth below-medan Tobn s q n a gven quarter. Subsamples n Columns (3) and (6) consst of banks wth above-medan Tobn s Q n a gven quarter. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely. 37

Table 9. Share of Mortgage-Backed Securtes that s Held-to-Maturty n 2008 Guaranteed Not guaranteed Guaranteed Not guaranteed Guaranteed Not guaranteed (1) (2) (3) (4) (5) (6) Loans 0.030-0.913*** 0.068-0.855*** 0.073-0.914*** (0.268) (0.284) (0.274) (0.269) (0.273) (0.292) Real estate loans 0.031 0.731*** 0.015 0.701*** -0.022 0.561** (0.208) (0.229) (0.207) (0.215) (0.218) (0.215) Real estate loans * Low valuaton 0.073 0.321 (0.155) (0.249) Securtes, amortzed cost 0.148-0.219 0.210-0.139 0.197-0.296 (0.244) (0.493) (0.253) (0.488) (0.245) (0.499) MBS, amortzed cost 0.354 1.048 0.332 1.036 0.303 0.649 (0.280) (0.685) (0.275) (0.680) (0.268) (0.695) MBS, amortzed cost * Low valuaton 0.098 1.196* (0.363) (0.662) Low valuaton 0.029 0.028-0.020-0.271 (0.019) (0.029) (0.103) (0.173) Large bank -0.019 0.060-0.016 0.063-0.016 0.066* (0.027) (0.041) (0.026) (0.040) (0.026) (0.039) HPI 0.017 0.022 0.017 0.021 0.016 0.024 (0.016) (0.034) (0.016) (0.034) (0.016) (0.032) Constant -0.101-0.028-0.148-0.069-0.121 0.068 (0.132) (0.296) (0.145) (0.297) (0.144) (0.305) 38 N 1098 582 1098 582 1098 582 R 2 0.237 0.434 0.241 0.436 0.242 0.451 The dependent varable s the share of mortgage-backed securtes that s held-to-maturty. Low valuaton s a dummy varable that takes a value of one f the bank has a Tobn s q less than one, and zero otherwse. See the Appendx for varable defntons and data sources. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely.

Table 10. Share of Non-Guaranteed Mortgage-Backed Securtes that s Held-to-Maturty n 2001-2007 2001 2002 2003 2004 2005 2006 2007 (1) (2) (3) (4) (5) (6) (7) Loans 0.116 0.293 0.029-0.169-0.641** -0.564** -0.619** (0.412) (0.363) (0.404) (0.234) (0.282) (0.281) (0.309) Real estate loans 0.137 0.037 0.138 0.346 0.533** 0.424** 0.475** (0.320) (0.286) (0.316) (0.212) (0.227) (0.206) (0.213) Securtes, amortzed 1.071** 1.367*** 0.781** 0.407 0.354 0.330 0.024 (0.508) (0.447) (0.368) (0.362) (0.361) (0.348) (0.321) MBS, amortzed -0.413-0.801-0.419 0.068 0.242-0.090 0.342 (0.604) (0.516) (0.378) (0.384) (0.469) (0.574) (0.683) Low valuaton 0.011 0.029 0.049 0.049-0.022-0.026-0.015 (0.076) (0.042) (0.035) (0.033) (0.037) (0.029) (0.035) Large bank -0.065-0.066-0.048-0.018 0.019 0.061 0.049 (0.090) (0.063) (0.054) (0.048) (0.047) (0.046) (0.043) HPI 0.114-0.048 0.043-0.099* 0.016 0.003 0.005 (0.095) (0.142) (0.115) (0.058) (0.048) (0.074) (0.026) Constant -0.574-0.281-0.428 0.460 0.065 0.238 0.047 (0.380) (0.376) (0.304) (0.364) (0.179) (0.216) (0.243) 39 N 126 522 516 507 581 572 556 R 2 0.381 0.346 0.270 0.316 0.349 0.274 0.438 The dependent varable s the share of mortgage-backed securtes that s held-to-maturty. Low valuaton s a dummy varable that takes a value of one f the bank has a Tobn s q less than one, and zero otherwse. See the Appendx for varable defntons and data sources. Regressons nclude state fxed effects and quarterly perod fxed effects (not reported). Data are based on quarterly observatons. Standard errors are corrected for clusterng at the bank level. *, **, and *** denote sgnfcance at the 10%, 5% and 1% levels, respectvely.

40 Fgure 1. Tobn s q and Share of Zombe Banks 1.12 1.10 1.08 1.06 1.04 1.02 1.00 0.98 0.96 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Tobn's Q (LHS) Zombe share (RHS) 0.94 0.0 2001Q4 2002Q4 2003Q4 2004Q4 2005Q4 2006Q4 2007Q4 2008Q4 Tobn s q s the rato of market value to book value of assets. Zombe share s the fracton of banks wth Tobn s q less than 1. Quarterly data from Call reports and Datastream. Fgure 2. Real Estate Loans and Mortgage-backed Securtes 0.56 0.54 0.14 0.12 0.52 0.50 0.48 0.46 0.44 0.42 2001Q4 2002Q4 2003Q4 2004Q4 2005Q4 2006Q4 2007Q4 2008Q4 0.10 0.08 0.06 0.04 0.02 Real estate loans (LHS) Mortgage-backed securtes (RHS) Real estate loans s the rato of real estate loans to total assets. Mortgage-backed securtes s the rato of MBS to total assets. Securtes are valued at amortzed cost f held-to-maturty and at far value f avalable-for-sale. Quarterly data from Call reports. 0.13 0.12 0.11 Fgure 3. Share of Mortgage-backed Securtes that s Held-to-Maturty 0.10 0.09 0.08 Guaranteed MBS (LHS) Non-guaranteed MBS (RHS) 0.07 0.06 0.05 2001Q4 2002Q4 2003Q4 2004Q4 2005Q4 2006Q4 2007Q4 2008Q4 Guaranteed MBS s the fracton of guaranteed MBS that s held-to-maturty. Non-guaranteed MBS s the fracton of nonguaranteed MBS that s held-to-maturty. Quarterly data from Call reports.

41 Fgure 4. Far Value of Mortgage-backed Securtes Relatve to Amortzed Cost 1.05 1.00 0.95 0.90 Guaranteed MBS (LHS) Non-guaranteed MBS (RHS) 0.85 0.80 2001Q4 2002Q4 2003Q4 2004Q4 2005Q4 2006Q4 2007Q4 2008Q4 Guaranteed MBS s the far value of guaranteed MBS to the amortzed value of guaranteed MBS. Non-guaranteed MBS s the far value of non-guaranteed MBS to the amortzed value of non-guaranteed MBS. Quarterly data from Call reports. Fgure 5. Captalzaton and Composton of Bank Regulatory Captal 0.13 0.89 0.12 0.88 0.11 0.10 0.87 0.86 Ter 1 captal to total assets (LHS) Ter 1 captal n total captal (RHS) 0.09 0.08 2001Q4 2002Q4 2003Q4 2004Q4 2005Q4 2006Q4 2007Q4 2008Q4 0.85 0.84 Ter 1 captal to total assets s the rato of ter 1 captal to total rsk-weghted assets. Ter 1 captal n total captal s the rato of ter 1 captal to total regulatory captal. Quarterly data from Call reports.